Domestic city gas supply began in the late 1980s and has been used for more than 30 years, and city gas is a very convenient and stable fuel on the premise of securing safety. However, most of the supply pipes of urban gas operators are buried in the soil, about 49.4% of which pipes are made of metal affected by corrosion.

It is a time that the importance of cathodic protection to City gas pipes that are used for long-term use. As a way to improve this, as artificial intelligence (AI, Artificial Intelligence) related technologies have recently advanced with the Fourth Industrial Revolution, the number of cases of AI grafting in various fields is increasing, and it is judged that this can be applied to the field of City gas pipes cathodic protection.

In this study, after collecting the cathodic protection potential data received from the remote type potential measurement terminal (T/B, Test Box) and the rectifier output voltage at the time, the AI control model let to learn. The learning data of the AI model was secured by data augmentation through regression analysis of the initial collected data, and the value-based Q-Learning model was applied among the Deep Reinforcement Learning (DRL) algorithms.

The AI control model that has completed data learning is put into the actual city gas supply area to verify that the AI responds appropriately based on the data from the received remote type potential measurement terminal, and through this, we tried to verify whether it can be used as an appropriate means for cathodic protection current AI control. I wanted to verify whether it could be done. Individual change characteristics were applied depending on the installation location, but since the potential data received through the remote potential measurement terminal reflects external factors that change in real time, it was very difficult to analyze this and control the anti-corrosion current. Individual change characteristics were applied according to the installation location, therefore it was very difficult to analyze and control the cathodic protection current because the potential data received through the remote potential measurement terminal reflects external factors that change in real time.
The reason is that the external environment, such as soil characteristics, temperature, and humidity, has different characteristics depending on the location of each installed potential measurement terminal. AI''s primary target setting range was set so that the cathodic protection potential of all potential measurement terminals settled within the range of D.C -1,600 ㎷ or more and -1,100 ㎷ or less. This is because the corrosion potential in that range was judged to be the most appropriate level for piping management.

By implementing the anti-corrosion current AI control system, results that could not be obtained with existing methods were obtained, such as constant provision of anti-corrosion potential, problem notification function, and reduction of rectifier output voltage. Although no results were obtained for all potential measurement terminals to be settled, we confirmed that the AI was continuously controlled to settle as many potential measurement terminals as possible within the primary target range, and verified the performance of the completed AI. The results obtained through the Boryeong area study showed that it was difficult to place all potential measurement terminals within the primary target range, considering the diverse environments of the buried area, such as the increase in pipes subject to corrosion and the complex urban structure. . This is because the initially set primary target range is a corrosion potential value that conservatively applies the maintenance standards for facilities subject to electrical corrosion, and it is difficult to settle it at the set value in all actual sites. Through this, when operating an electrical corrosion management system using AI, it is judged necessary to separate and set the primary target range for electrical corrosion potential by considering the characteristics of each region. Based on the above results, this study concluded that in order to prevent corrosion of city gas pipes, the introduction of a corrosion protection current AI control system is necessary under the following conditions.

In addition, some potential measurement terminals have introduced a data filter function, indicating that the system needs to be supplemented so that AI can perform reliable rectifiers. In addition to the data filter function, it was necessary to increase the number of potential measurement terminals. This is because terminal data for potential measurement, which is the most important factor for AI to determine rectifier control, must be sufficiently secured when filtering data. Studies have shown that increasing the number of terminals for remote type potential measurement brings higher reliability to AI''s control of electric rectifiers. Based on the above results, this study concluded that in order to prevent corrosion of city gas pipes, it is necessary to introduce an electric rectifier AI control system under the following conditions.

First, as seen in some corrosion accident cases, it can be seen that city gas pipes are quite vulnerable to time-dependent threats when used for more than about 20 years. Therefore, it suggests the need to introduce AI systems in pipes that will be used for more than 20 years. The reason is that, considering the coating damage rate of about 2%, the possibility of surface corrosion due to long-term use of the area and the vulnerability due to long-term use of the adhesive used in manufacturing polyethylene clad steel pipes (PLP), which are mainly used as city gas pipes, etc. exist.

Second, according to environmental characteristics such as high soil resistivity and frequent excavation work, a potential measurement terminal should be installed in an area where city gas pipes are exposed to the corrosion environment, so that the provision of cathodic protection potential can be provided stably at all times.

As confirmed by the research, it is expected that AI will replace and manage many of the existing cathodic protection tasks that humans have been doing, which will greatly contribute to improving the soundness of city gas piping, such as improving the life of rectifiers and stabilizing the cathodic protection potential.